Mozilla Releases DeepSpeech 0.7 As Their Great Speech-To-Text Engine

[starshipeleven]

Do you want to world stalled for ever?

No.

No one can stop the flow of the time.

I know, Skynet has always been inevitable. I don't see why you are so upset.

[starshipeleven]

Angry mobs don't need proof. You pay them 50$ each and they will do in to the streets to destroy, loot and kill. This is how the real world works.

No that's not how the world works, manipulating stupid people does not work as well if you are just paying them off.

That's why there are agents and bots alike, spreading bs on social media.

People that "destroy, loot and kill" in anywhere near an effective manner aren't going to cost you 50$ each. For that price you get drunkards and junkies.

[pabloski]

Do you want Skynet? Because that's how you get Skynet.

Do you want to world stalled for ever? Sorry, this ain't happen. No one can stop the flow of the time. Are you one of those people who would have killed Henry Ford just to stop the "car invasion"?

[pabloski]

Photomontages aren't near-realtime, and forensic experts might not be able to save you from an angry mob, or undo massive political damage done by an impersonator with this technology.

Angry mobs don't need proof. You pay them 50$ each and they will do in to the streets to destroy, loot and kill. This is how the real world works.

[gregzeng]

Old Timers like myself have paid literally thousands of dollars for "Dragon Naturally Speaking", in Windows & other operating systems. There have been many imitators to DNS. Then these imitators improved so much that they almost replace DNS. The commercial prices have dropped. The quality, power & flexibility is now so high.
This open source iniative from Mozilla is a big test of Private-Selfish-Commercialism, versus Open-Source.
In between these two extremes are many models of business & government. Both Mozilla & Microsoft etc has joined OIN, The Open Invention Network, which is good for Linux. Perhaps some of Microsoft's 90,000 patents might be useful for future Speech To Text developments.

[starshipeleven]

Then you build a more powerful and astute machine to stop the fakes.

Do you want Skynet? Because that's how you get Skynet.

[starshipeleven]

Same with photomontage. And like the only technologies, this new tech leaves artifacts, that forensic expert are able to identify to say if it say a fake or true voice. And the game of cat and mouse goes on and on.

Photomontages aren't near-realtime, and forensic experts might not be able to save you from an angry mob, or undo massive political damage done by an impersonator with this technology.

[pabloski]

the machine that spots the fakes no longer can spot them anymore. It is scary.

Then you build a more powerful and astute machine to stop the fakes. NN models evolve with time and governments have the manpower and the financial resources to develop new models and run them on uber powerful hardware.

I can give you an example from my "professional" life. I built ( many years ago ) a sophisticated Markov chain model to write blog articles to fool Google. It worked for a couple years. Then they developed a new semantic engine, based on neural networks and blower out of the water the blogs writter by my, less powerful and astute, Markov model.

[uid313]

Same with photomontage. And like the only technologies, this new tech leaves artifacts, that forensic expert are able to identify to say if it say a fake or true voice. And the game of cat and mouse goes on and on.

Yeah, but then you can create a machine that creates fake, and a machine that spots fakes, then you feed each machine with the output of the other until the fake is so good that the machine that spots the fakes no longer can spot them anymore. It is scary.

[tildearrow]

It is pretty scary what you can do with machine learning. With a 5 second audio clip you feed that to a machine learning software and clone that voice and make it say whatever you want. So you can make it sound like the president, or your friend, or if you have an enemy, it can make it sound like your enemy. Someone could make it sound like you and and make it say that it admits to murdering someone.



It is not a voice recognition engine. It is a text-to-speech engine.

DeepSpeech is a voice recognition engine.
I can't believe 3 people believe you.

Read the title. "Speech-to-Text".